imcc/docs/syntax.pod has the following:
=item 'char constant'
Are delimited by B<'>. They are taken to be C<ascii> encoded. No escape
sequences are processed.
But in fact B<'> behaves like B<"> without escapes or encoding/charset
prefixes:
[EMAIL PROTECTED]> cat foo.pir
.sub _main @MAIN
S1 = 'xyzzy\n'
print S1
print ascii:"\n"
.end
[EMAIL PROTECTED]> parrot foo.pir
xyzzy\n
[EMAIL PROTECTED]>
A syntax for specifying multiple characters without escapes seems like a
useful thing, a la Perl5, but being unable to specify an encoding or
charset seems less useful, even for a single character. In fact, I was
expecting a single character to be represented as an integer internally,
or at least to be easily converted to one, so I was surprised that this
didn't print "120\n":
[EMAIL PROTECTED]> cat bar.pir
.sub _main @MAIN
I1 = 'x'
print I1
print "\n"
.end
[EMAIL PROTECTED]> parrot bar.pir
0
[EMAIL PROTECTED]>
That would seem to be the one reasonable use for character constants,
but 'x' behaves no differently from "x" in this example.
So what should 'x' mean in PIR? I would suggest:
1. B<'> in PIR is like B<'> in Perl5, i.e. accept the status quo,
but add encoding/charset prefix syntax, and fix the doc; and
2. Support character constants via either string syntax by defining
I1 = 'x'
mean the equivalent of
S1 = 'x'
I1 = ord S1
to match my naive assumption. Does that sound reasonable?
-- Bob Rogers
http://rgrjr.dyndns.org/
